Power device life prediction method and system based on multi-physics field coupling

Abstract

This invention belongs to the field of power electronic device technology, specifically relating to a method and system for predicting the lifetime of power devices based on multi-physics coupling. The prediction method includes: load data acquisition and intelligent compression; dynamic fitting of a multi-factor lifetime model; rainflow counting and matrix aggregation; cumulative damage calculation and lifetime prediction; and Weibull statistical analysis. By constructing a multi-factor dynamic correction model, multi-physics parameters such as conduction time ton, load current I, blocking voltage V, and bond wire diameter D can be incorporated into regression analysis. The system allows users to flexibly select influencing factors based on experimental data and automatically solves the nonlinear correction coefficients based on the least squares method, thereby achieving accurate lifetime assessment of power devices under electro-thermal-mechanical multi-field coupling stress. Furthermore, addressing the problem of large amounts of long-period measured load spectrum data and time-consuming calculations, this invention innovatively introduces a rainflow matrix aggregation mechanism. By mapping the extracted massive amounts of heat dissipation cycles onto a preset two-dimensional statistical grid, millions of cycle data are compressed into a fixed-dimensional matrix for batch calculation while preserving the accuracy of damage characteristics.

Description

Technical Field

[0001] This invention belongs to the field of power electronic device technology, specifically relating to a method and system for predicting the lifetime of power devices based on multi-physics coupling. Background Technology

[0002] With the widespread application of power electronics technology in new energy vehicles, wind power generation, rail transportation, and aerospace, power semiconductor devices (such as IGBTs and MOSFETs) are core components for energy conversion, and their reliability directly affects the safety and stable operation of the entire system. In actual operating conditions, power devices often endure complex and variable thermomechanical stresses. Due to the mismatch in the coefficients of thermal expansion (CTE) of materials such as the chip, solder layer, and substrate, cumulative fatigue damage can occur inside the device under cyclic temperature changes (thermal cycling), ultimately leading to failure modes such as bond wire detachment or solder layer aging and delamination.

[0003] To assess the remaining lifetime of power devices, the industry typically employs a physical-of-failure (PoF) approach. The general procedure for this approach includes: 1. Load Acquisition: Obtain the junction temperature T0 of the device under a specific mission profile through actual measurement or simulation. j Time series data; 2. Rainflow Counting: The Rainflow Counting Algorithm is used to transform irregular temperature fluctuations into a series of closed stress cycles, and the junction temperature fluctuation amplitude ΔT of each cycle is extracted. j Average junction temperature Tm and number of cycles; 3. Damage calculation: The damage caused by each stress cycle is calculated based on classical life prediction models (such as the Coffin-Manson model, the Arrhenius model, or their modified models); 4. Linear accumulation: According to Miner's linear accumulation damage theory, the damage of all cycles is superimposed. When the total damage reaches the threshold (usually 1), the device is determined to be faulty.

[0004] In addition, in order to obtain more than just a single lifetime value, it is also necessary to combine statistical methods (such as Weibull distribution) to analyze the reliability indicators of the device under different failure probabilities (such as B10 lifetime, MTTF, etc.), so as to provide a basis for the product's operation and maintenance strategy.

[0005] Although the aforementioned physical model-based lifetime prediction methods have been applied in engineering, existing technical solutions and software tools still have the following obvious defects and shortcomings when processing large-scale measured data and multiphysics coupling analysis: 1. Life prediction models have fixed parameters and lack the ability to dynamically adjust for multiple factors. Existing commercial lifetime prediction software or general calculation methods typically incorporate fixed empirical formulas (such as the standard LESIT model). Most of these models only consider the junction temperature difference ΔT. j and average junction temperature T m Two main variables. However, in practical applications, the device lifetime is also significantly affected by the on-time t. on The influence of process and electrical parameters such as load current I, blocking voltage V, and even bond wire diameter D is significant. Existing technologies often cannot flexibly incorporate these parameters as variables into regression analysis, or require users to manually derive complex correction coefficients. This results in insufficient prediction accuracy of the model under specific complex operating conditions, making it unable to adapt to the customized fitting needs of new devices or special application scenarios.

[0006] 2. Rainflow counting is inefficient with large datasets and lacks a data aggregation mechanism. In long-term task profile analysis (such as the entire lifecycle driving data of electric vehicles), the number of sampling points often reaches millions or even tens of millions, resulting in an extremely large number of extracted rainflow cycles. Current technologies typically process all discrete cycle data individually, consuming enormous memory resources and significantly reducing computational speed. This also makes visualization of damage distribution difficult. There is a lack of an efficient "matrix aggregation" or "binning" mechanism—that is, a mechanism that can automatically merge massive discrete cycles into a finite two-dimensional grid (Range-Mean Matrix) for accelerated computation and storage while maintaining accuracy.

[0007] 3. The physical model and statistical analysis tools are disconnected, lacking an integrated evaluation process. Current life assessment processes are typically fragmented: engineers first need to perform rainflow counting and damage calculations in one tool to obtain the average lifespan, and then export the data to another statistical software (such as Minitab) for Weibull distribution fitting to assess the failure probability. This step-by-step operation is not only inefficient and prone to data transmission errors, but also makes it difficult to achieve real-time linkage analysis from "model parameter adjustment" to "reliability probability change". Existing technologies lack a full-process expert system that integrates experimental data fitting, load spectrum rainflow analysis, and Weibull reliability assessment, and cannot intuitively demonstrate the sensitivity impact of parameter changes on the final B10 or B50 lifespan.

[0008] 4. Lack of intuitive visualization methods for damage distribution Existing tools often only output the final lifespan value (e.g., "15.3 years"), without visually indicating which operating condition (e.g., large temperature difference, low frequency cycling, or small temperature difference, high frequency cycling) caused the main damage to the lifespan. The lack of a visual analysis method that links the rainflow counting matrix (thermograph) with the lifespan model curve on the same screen makes it difficult for engineers to optimize heat dissipation design or control strategies in a targeted manner. Summary of the Invention

[0009] The purpose of this invention is to provide a method, system, electronic device, and storage medium for predicting the lifetime of power devices based on multi-physics coupling.

[0010] The first aspect of this application provides a power device lifetime prediction method based on multi-physics coupling, comprising: Payload data acquisition and intelligent compression; Dynamic fitting of multi-factor lifetime models; Rainflow counting and matrix aggregation; Cumulative damage calculation and lifespan prediction; Weibull statistical analysis.

[0011] In one embodiment of this application, the method for acquiring and intelligently compressing payload data includes: Receives data including time series and junction temperature T. j The raw payload data of the sequence; Data compression is performed using a compression algorithm based on the preservation of extreme value features.

[0012] In one embodiment of this application, the method for data compression using a compression algorithm based on extreme value feature preservation includes: Extreme value extraction: Using the relative extreme value algorithm, identify all local maxima and local minima in the temperature sequence; Force the preservation of the start and end points of the time series; The extracted maxima, minima, and endpoints are reassembled in chronological order to form a compressed characteristic load spectrum.

[0013] In one embodiment of this application, the method for dynamic fitting of the multi-factor lifetime model includes: Establish a general nonlinear life model that incorporates multiple stress factors: ; Solve for the parameters using the following steps: Factor selection: Obtain the influencing factors selected by the user through the interactive interface to participate in the fitting; Linearization: Taking the natural logarithm of both sides of the nonlinear lifetime model equation transforms it into a multiple linear regression form. ; Least squares solution: Construct the feature matrix X and the target vector Y, and use the least squares method to solve for the regression coefficient vector [ln(K), β1,β2,......]; Parameter output: The solved regression coefficients are converted back into physical model parameters to obtain the lifetime prediction formula for a specific device.

[0014] In one embodiment of this application, the method for rainflow counting and matrix aggregation includes: Cyclic Extraction: The temperature sequence is processed using the standard rainflow counting algorithm to extract several thermal cycles. Each cycle contains a feature tuple: Range ΔT Mean Tj ,Count; Matrix-based aggregation: Set the number of segments N bins On the ΔT axis and T m Divide the axis into N bins Construct N uniform intervals. bins *N bins Two-dimensional grid; Iterate through all extracted discrete loops and map them to the corresponding grid cells based on their magnitude and mean. The number of cycles falling into the same grid cell is accumulated, and the weighted average magnitude and weighted average mean of all cycles within that cell are calculated.

[0015] In one embodiment of this application, the method for cumulative damage calculation and lifetime estimation includes: Traverse each non-empty grid cell i in the aggregated rainflow matrix; Extract the weighted average magnitude ΔT of the grid cell. i and mean T mi By combining the fixed parameters set for the current operating condition with the trained nonlinear life model, the cycle life N under this operating condition is calculated. f,i ; Calculate the unit damage Di= , where n i This represents the total number of iterations within the grid. The total damage is obtained by summing all mesh damage: ; Life expectancy is calculated based on the total duration (Tduration): Life = .

[0016] In one embodiment of this application, the Weibull statistical analysis method includes: Construct the likelihood function of the two-parameter Weibull distribution; Iteratively solve for the shape parameter β and the scale parameter η; Calculate the cumulative failure probability function: ; Key reliability indicators are solved in reverse.

[0017] A second aspect of the present invention provides a power device lifetime prediction system based on multiphysics coupling, comprising: The data acquisition and compression module is used for payload data acquisition and intelligent compression. The model training and parameter fitting module is used for dynamic fitting of multi-factor lifetime models. Matrix aggregation module, used for rainflow counting and matrix aggregation; The lifespan prediction module is used for cumulative damage calculation and lifespan prediction. The analysis module is used for Weibull statistical analysis.

[0018] A third aspect of the present invention provides an electronic device including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the power device lifetime prediction method based on multiphysics coupling as described above.

[0019] A fourth aspect of the present invention provides a storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device on which the storage medium is located to perform the power device lifetime prediction method based on multiphysics coupling as described above.

[0020] The beneficial effects of this invention are: This invention constructs a multi-factor dynamic correction model, which can adjust the conduction time t. on The system incorporates multiple physical field parameters, such as load current I, blocking voltage V, and bond wire diameter D, into the regression analysis. Users can flexibly select influencing factors based on experimental data, and the system automatically solves for nonlinear correction coefficients using the least squares method, thereby achieving accurate lifetime assessment of power devices under electro-thermal-mechanical multi-field coupled stress. Furthermore, addressing the issues of large amounts of long-period measured load spectrum data and time-consuming calculations, this invention innovatively introduces a rainflow matrix aggregation mechanism. By mapping the extracted massive amounts of heat dissipation cycles onto a preset two-dimensional statistical grid, millions of cycle data points are compressed into fixed-dimensional matrices (such as 50*50 grids) for batch calculations while preserving the accuracy of damage characteristics (amplitude and mean).

[0021] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0023] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of a preferred embodiment of the power device lifetime prediction method based on multi-physics coupling of the present invention; Figure 2 This is a flowchart of the overall architecture of a power device lifetime prediction system based on multi-physics coupling, according to a preferred embodiment of the present invention. Figure 3 This is a schematic diagram of the human-computer interaction interface of the "model training and fitting" module in one embodiment of the present invention; Figure 4 This is a schematic diagram of a dual-coordinate system prediction curve generated by the "prediction visualization analysis" module in one embodiment of the present invention; Figure 5 This is a PDF probability density and CDF cumulative failure probability map generated by the "Weibull distribution analysis" module in one embodiment of the present invention; Figure 6 This is a schematic diagram of the integrated analysis dashboard of the "Rainflow Counting and Lifetime Assessment" module in one embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] This application provides a method, system, electronic device, and storage medium for predicting the lifetime of power devices based on multiphysics coupling, which are described in detail below. It should be noted that the order of description of the following embodiments is not intended to limit the preferred order of the embodiments of this application. Furthermore, the descriptions of each embodiment have their own emphasis; parts not described in detail in a certain embodiment can be referred to in the relevant descriptions of other embodiments.

[0027] See Figure 1 In one embodiment, the power device lifetime prediction method based on multiphysics coupling includes: S1, Load data acquisition and intelligent compression; S2, dynamic fitting of multi-factor lifetime model; S3, Rainflow counting and matrix aggregation; S4, Cumulative damage calculation and lifetime prediction; S5, Weibull statistical analysis.

[0028] Optionally, in step S1, the method for acquiring and intelligently compressing the load data includes: Receives data including time series and junction temperature T. j The original load data of the sequence (CSV / Excel format); considering that the amount of measured data may be huge (millions of sampling points), in order to reduce the amount of calculation without losing the peak and trough information that is crucial to fatigue damage, a compression algorithm based on the preservation of extreme value features can be used for data compression.

[0029] Specifically, the method for data compression using a compression algorithm based on extreme value feature preservation includes: Extreme value extraction: Using the relative extreme value algorithm (argrelextrema), all local maxima and local minima in the temperature sequence are identified; Force the preservation of the start and end points of the time series; The extracted maxima, minima, and endpoints are reassembled in chronological order to form a compressed characteristic load spectrum.

[0030] In this embodiment, compared to simple equal-interval downsampling, this method ensures the accuracy of the cyclic amplitude in subsequent rainflow counting, while significantly reducing the data dimensionality.

[0031] Furthermore, in step S2, the method for dynamically fitting the multi-factor lifetime model includes: Establish a general nonlinear life model that incorporates multiple stress factors: ; Solve for the parameters using the following steps: Factor selection: Obtain the influencing factors selected by the user through the interactive interface to participate in the fitting; Linearization: Taking the natural logarithm of both sides of the nonlinear lifetime model equation transforms it into a multiple linear regression form. ; Least squares solution: Construct the feature matrix X and the target vector Y, and use the least squares method to solve for the regression coefficient vector [ln(K), β1,β2,......]; Parameter output: The solved regression coefficients are converted back into physical model parameters to obtain the lifetime prediction formula for a specific device.

[0032] Furthermore, in step S3, the method for rainflow counting and matrix aggregation includes: Cyclic Extraction: The temperature sequence is processed using the standard Rainflow Counting Algorithm to extract several thermal cycles. Each cycle contains a feature tuple: Range ΔT Mean Tj ,Count; Binning Strategy: Set the number of segments N bins (e.g., 50), on the ΔT axis and T m Divide the axis into N bins Construct N uniform intervals. bins *N bins Two-dimensional grid; Iterate through all extracted discrete loops and map them to the corresponding grid cells based on their magnitude and mean. The number of cycles falling into the same grid cell is accumulated, and the weighted average magnitude and weighted average mean of all cycles within that cell are calculated.

[0033] In this embodiment, the original circular list of length M (which may be in the millions) can be converted into a statistical matrix of fixed size (such as 50*50), which greatly improves the efficiency of subsequent damage calculation and facilitates the generation of heat maps.

[0034] Furthermore, in step S4, the method for calculating cumulative damage and estimating lifespan includes: Traverse each non-empty grid cell i in the aggregated rainflow matrix; Extract the weighted average magnitude ΔT of the grid cell. i and mean T mi Combined with the fixed parameters set under the current operating conditions (such as I, V, t) on Substitute the values ​​into the trained nonlinear lifetime model to calculate the cycle life N under this operating condition. f,i ; Calculate the unit damage Di= , where n i This represents the total number of iterations within the grid. The total damage is obtained by summing all mesh damage: ; Life expectancy is calculated based on the total duration (Tduration): Life = .

[0035] Furthermore, in step S5, the maximum likelihood estimation (MLE) method can be used to perform statistical analysis on multiple sets of lifetime test data. Specifically, the Weibull statistical analysis method includes: Construct the likelihood function of the two-parameter Weibull distribution; Iteratively solve for the shape parameter β and the scale parameter η; Calculate the cumulative failure probability function: ; Reverse engineering key reliability metrics, such as B10 lifetime (lifetime with a 10% failure probability) and B50 lifetime.

[0036] It is understandable that the order of steps S1 to S5 can be adjusted as needed.

[0037] Accordingly, one embodiment of this application also provides a power device lifetime prediction system based on multiphysics coupling, comprising: The data acquisition and compression module is used for payload data acquisition and intelligent compression. The model training and parameter fitting module is used for dynamic fitting of multi-factor lifetime models. Matrix aggregation module, used for rainflow counting and matrix aggregation; The lifespan prediction module is used for cumulative damage calculation and lifespan prediction. The analysis module is used for Weibull statistical analysis.

[0038] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0039] Those skilled in the art will clearly understand that, for convenience and brevity, the specific working process of the system described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0040] Furthermore, one embodiment of this application provides an electronic device including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the power device lifetime prediction method based on multiphysics coupling as described above.

[0041] The electronic device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The electronic device may include, but is not limited to, a processor and a memory.

[0042] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0043] The memory can be used to store the computer program. The processor implements various functions of the electronic device by running or executing the computer program stored in the memory and calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created based on the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0044] Furthermore, one embodiment of this application provides a storage medium, characterized in that the storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device where the storage medium is located to execute the power device lifetime prediction method based on multiphysics coupling as described above. The computer program includes computer program code, which may be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0045] In a specific application scenario, see Figure 2 This is an overall flowchart of a power device lifetime prediction system based on multiphysics coupling, according to one embodiment.

[0046] System Hardware and Environment: The power device lifetime prediction system based on multiphysics coupling provided in this embodiment of the invention can run on general-purpose computer equipment. The software is developed using Python, with a graphical user interface (GUI) built on the PyQt6 framework. Numerical calculations are performed using NumPy and Pandas, statistical analysis is performed using SciPy, and charts are drawn using Matplotlib.

[0047] Specifically, the system can mainly include four functional modules, which correspond to the four tabs of the software interface.

[0048] Part 1: Training and Parameter Fitting of Multiphysics Models (see...) Figure 3 ) Users first select the physical factors affecting lifespan on the left side of the interface, including: junction temperature difference ΔT j Minimum junction temperature T jmin On-time t on Load current I, blocking voltage V, and bonding wire diameter D.

[0049] Subsequently, the user entered multiple sets of accelerated aging test (ALT) data into the table. Each row of data included the specific values ​​of the aforementioned physical factors and the corresponding failure lifetime N. f .

[0050] In response to the user's "Start Fit Analysis" command, the system performs the following steps: 1. Data linearization: Transforming nonlinear physical equations into linear regression forms. For T... jminThe Arrhenius model (1 / T) is used, and the Coffin-Manson power law form ln(x) is used for other variables.

[0051] 2. Least squares solution: Call the DynamicFitter.fit method to calculate the regression coefficients (β values) of each factor and the constant term K.

[0052] 3. Results Feedback: The goodness-of-fit R-value is displayed in real time on the interface. 2 The coefficient values ​​of each parameter are provided for users to evaluate the accuracy of the model.

[0053] Part Two: Visual Analysis of Predictive Models (see...) Figure 4 ) This section corresponds to the `setup_plot_tab` and `MplCanvas.update_plots` methods in the code.

[0054] To visually demonstrate the sensitivity of each parameter to the lifespan, the system provides an interactive plotting function.

[0055] 1. Variable control: The user selects a main variable from the drop-down menu (e.g., ΔT for the X-axis). j The system automatically locks other variables (such as I, V, t). on () is a fixed value input box.

[0056] 2. Curve generation: The system calculates the variation trend of lifetime Nf within the X-axis range based on the model parameters trained in the first part.

[0057] 3. Dual-view display: such as Figure 4 As shown, the system simultaneously draws two subgraphs: Linear coordinate graph: Displays an intuitive trend of declining lifespan.

[0058] Double logarithmic coordinate graph: Displays the linear trend under the power law relationship, which is convenient for verifying the physical consistency of the model.

[0059] And dynamically generate and display the corresponding mathematical formulas (LaTeX format) in the chart, for example: .

[0060] Part Three: Reliability Statistical Analysis (see...) Figure 5 ) This section corresponds to the WeibullAnalyzer class in the code.

[0061] The system not only focuses on average lifetime but also assesses statistical probability lifetime. The user inputs a set of lifetime test data for devices in the same batch, and the system uses maximum likelihood estimation (MLE) to fit a two-parameter Weibull distribution.

[0062] like Figure 5 As shown, the system generates two charts: The above figure (PDF) shows the probability density function curve and automatically labels the characteristic lifetime (η, Mean) and B5 lifetime point.

[0063] The following figure (CDF) shows the cumulative failure probability curve, allowing users to intuitively read the B10 lifetime value corresponding to any probability (such as 10%).

[0064] Part Four: Rainflow Counting and Measured Load Assessment (see...) Figure 6 ) This section is the core embodiment of the present invention, corresponding to Tab 4, the RainflowService class, and the DataLoadWorker class in the code.

[0065] Step S4-1: Intelligent import and compression of measured data Users import measured load spectra (CSV / Excel) containing timestamps and junction temperature data. Since the raw data may contain millions of sampling points, the DataLoadWorker module preprocesses the data using an argrelextrema algorithm, retaining only peaks, troughs, and endpoints while discarding invalid intermediate data. This significantly reduces memory usage while maintaining the accuracy of rainflow counting.

[0066] Step S4-2: Rainflow counting and matrix aggregation The system calls the RainflowService.aggregate_cycles method to perform the following operations: 1. Extracting cycles: Extract all thermal cycles from the compressed waveform and obtain the amplitude (Range) and mean (Mean) of each cycle.

[0067] 2. Grid division: Based on the number of segments set by the user (num_bins, for example, 50 segments), the "amplitude-mean" plane is divided into a 50*50 grid.

[0068] 3. Aggregate statistics: Map all discrete cycles to the corresponding grids and calculate the cumulative number of cycles (Count Sum) within each grid.

[0069] Specific effect of this implementation: This step transforms millions of cyclic data points into a compact sparse matrix, greatly improving the speed of subsequent damage calculations.

[0070] Step S4-3: Damage Calculation and Integrated Dashboard Display The system calculates the total damage based on Miner's linear cumulative damage theory and displays a comprehensive dashboard on the right side of the interface: Area 1 (top): Displays the time-domain waveform of the measured junction temperature, providing a visual representation of temperature fluctuations.

[0071] Area 2 (bottom left): Displays the model curve ΔTj of the current application, indicating the baseline for damage calculation.

[0072] Region 3 (bottom right - core features): Displays the rainflow counting matrix heatmap.

[0073] The horizontal axis represents the average junction temperature (Mean temp), and the vertical axis represents the temperature difference amplitude (ΔT).

[0074] The shade of color represents the number of cycles that occur under that operating condition or the weight of the damage caused.

[0075] The heat map directly reveals the "culprit" operating conditions that cause device failure (for example, the darkest area in the figure corresponds to the area with large temperature difference and high frequency).

[0076] Through the coordinated work of the above four parts, a closed-loop prediction system from data fitting to final measured load evaluation was realized.

[0077] In summary, compared with the prior art, the present invention has the following significant advantages and beneficial effects: 1. Significantly improved the accuracy and adaptability of life prediction under complex operating conditions. Existing technologies typically use fixed empirical formulas (such as considering only temperature difference ΔTj and mean value Tm), which are difficult to adapt to new devices or special operating conditions. This invention constructs a multi-factor dynamic correction model through the DynamicFitter module, which can incorporate multiple physical field parameters such as conduction time ton, load current I, blocking voltage V, and bond wire diameter D into regression analysis. The system allows users to flexibly select influencing factors based on experimental data and automatically solves for nonlinear correction coefficients using the least squares method, thereby achieving accurate lifetime assessment of power devices under electro-thermal-mechanical multi-field coupled stress.

[0078] 2. Solved the technical problem of low efficiency in processing massive payload data. To address the issues of large data volumes and time-consuming computation of long-period measured load spectrum data, this invention innovatively introduces a rainflow matrix aggregation mechanism into RainflowService. By mapping the extracted massive amounts of heat dissipation cycles onto a preset two-dimensional statistical grid (binning), millions of cycle data points are compressed into fixed-dimensional matrices (e.g., 50*50 grids) for batch computation while preserving the accuracy of damage features (amplitude and mean). This not only significantly reduces memory usage but also improves the computation speed of damage assessment by several orders of magnitude.

[0079] 3. Intelligent data compression with waveform feature preservation was achieved. This invention incorporates an intelligent compression algorithm within the data loading stage (DataLoadWorker). Utilizing the principle of relative extrema (argrelextrema), it automatically identifies and retains all peaks, troughs, and endpoints in the temperature waveform, discarding only intermediate transition points that do not contribute to fatigue damage. This method effectively solves the problem of traditional downsampling methods easily losing small-amplitude cycles or causing underestimation of damage due to peak-shaving and valley-filling, significantly improving the system's response speed while ensuring data integrity.

[0080] 4. Provides an integrated expert system that deeply integrates physical models and statistical analysis. This invention breaks down the traditional disconnect between "physical modeling" and "reliability statistics." The system integrates a complete workflow, from experimental data fitting (Tab 1), prediction visualization (Tab 2), Weibull distribution analysis (Tab 3), to rainflow counting and damage assessment (Tab 4). This integrated design allows adjustments to model parameters to be fed back into the final lifetime prediction results in real time, significantly reducing human error caused by data transfer between different software programs and improving engineering analysis efficiency.

[0081] 5. A multi-dimensional, interconnected damage visualization and analysis dashboard was constructed. This invention achieves simultaneous display of time-domain waveforms, lifetime model curves, and rainflow count heatmaps through a customized plot layout (plot_rainflow_dashboard). A two-dimensional histogram visually illustrates the cycle frequency distribution under different combinations of temperature differences and mean values, helping engineers quickly locate "hot spots" causing major device damage and providing intuitive data support for heat dissipation design optimization and control strategy improvement.

[0082] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for predicting the lifetime of power devices based on multiphysics coupling, characterized in that, include: Payload data acquisition and intelligent compression; Dynamic fitting of multi-factor lifetime models; Rainflow counting and matrix aggregation; Cumulative damage calculation and lifespan prediction; Weibull statistical analysis.

2. The power device lifetime prediction method according to claim 1, characterized in that, The method for acquiring and intelligently compressing payload data includes: Receives data including time series and junction temperature T. j The raw payload data of the sequence; Data compression is performed using a compression algorithm based on the preservation of extreme value features.

3. The power device lifetime prediction method according to claim 2, characterized in that, The method for data compression using a compression algorithm based on extreme value feature preservation includes: Extreme value extraction: Using the relative extreme value algorithm, identify all local maxima and local minima in the temperature sequence; Force the preservation of the start and end points of the time series; The extracted maxima, minima, and endpoints are reassembled in chronological order to form a compressed characteristic load spectrum.

4. The power device lifetime prediction method according to claim 1, characterized in that, The method for dynamic fitting of the multi-factor lifetime model includes: Establish a general nonlinear life model that incorporates multiple stress factors: ； Solve for the parameters using the following steps: Factor selection: Obtain the influencing factors selected by the user through the interactive interface to participate in the fitting; Linearization: Taking the natural logarithm of both sides of the nonlinear lifetime model equation transforms it into a multiple linear regression form. ； Least squares solution: Construct the feature matrix X and the target vector Y, and use the least squares method to solve for the regression coefficient vector [ln(K), β1,β2,......]; Parameter output: The solved regression coefficients are converted back into physical model parameters to obtain the lifetime prediction formula for a specific device.

5. The power device lifetime prediction method according to claim 1, characterized in that, The method for rainflow counting and matrix aggregation includes: Cyclic Extraction: The temperature sequence is processed using the standard rainflow counting algorithm to extract several thermal cycles. Each cycle contains a feature tuple: Range ΔT Mean Tj ,Count; Matrix-based aggregation: Set the number of segments N bins On the ΔT axis and T m Divide the axis into N bins Construct N uniform intervals. bins *N bins Two-dimensional grid; Iterate through all extracted discrete loops and map them to the corresponding grid cells based on their magnitude and mean. The number of cycles falling into the same grid cell is accumulated, and the weighted average magnitude and weighted average mean of all cycles within that cell are calculated.

6. The power device lifetime prediction method according to claim 1, characterized in that, The methods for calculating cumulative damage and predicting lifespan include: Traverse each non-empty grid cell i in the aggregated rainflow matrix; Extract the weighted average magnitude ΔT of the grid cell. i and mean T mi By combining the fixed parameters set for the current operating condition with the trained nonlinear life model, the cycle life N under this operating condition is calculated. f,i ; Calculate the unit damage Di= , where n i This represents the total number of iterations within the grid. The total damage is obtained by summing all mesh damage: ; Life expectancy is calculated based on the total duration (Tduration): Life = .

7. The power device lifetime prediction method according to claim 1, characterized in that, The methods used in the Weibull statistical analysis include: Construct the likelihood function of the two-parameter Weibull distribution; Iteratively solve for the shape parameter β and the scale parameter η; Calculate the cumulative failure probability function: ; Key reliability indicators are solved in reverse.

8. A power device lifetime prediction system based on multiphysics coupling, characterized in that, include: The data acquisition and compression module is used for payload data acquisition and intelligent compression. The model training and parameter fitting module is used for dynamic fitting of multi-factor lifetime models. Matrix aggregation module, used for rainflow counting and matrix aggregation; The lifespan prediction module is used for cumulative damage calculation and lifespan prediction. The analysis module is used for Weibull statistical analysis.

9. An electronic device, characterized in that, The device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the power device lifetime prediction method based on multiphysics coupling as described in any one of claims 1-7.

10. A storage medium, characterized in that, The storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device where the storage medium is located to perform the power device lifetime prediction method based on multiphysics coupling as described in any one of claims 1-7.

Images